Apple sues OpenAI, public betas, & super apps on the AppleInsider Podcast

**William Gallagher** (0:04)
Hello, welcome to the Apple Insider Podcast. I'm William Gallagher. Our sponsor this week is CleanMyMac by MacPaw. More about that later. Well, unless that is, you're a subscriber to Apple Insider Plus, which we'll talk about more in a second. So much to talk about this time, but starting with how I am joined as ever by Wesley Hilliard. And Wes, you're back from breaks and holidays and things, and you've come back just as lawsuits have started between Apple and OpenAI. I thought they were friends. What's going on there?

**Wesley Hilliard** (0:36)
Well, the thing is, I feel like it was an uneasy partnership in the first place, right? So if we go back in time to 2023, when the whole world is like Apple, you're failing at life, you're not ready for AI, you're not ready for anything, you're doing bad, whatever.
Apple's trying to develop, well, it was already developing. We knew that Apple has been developing anything related to AI since 2016 The Apple car, you can trace Apple Intelligence, AI, Apple's version of it, back to 2014 with Project Titan, right? So it's not as if-

**William Gallagher** (1:18)
Actually, you might even be able to go even further. When was Snow Leopard? Because remember, in the middle of Snow Leopard, they actually say on stage that something or other isn't AI feature. So yeah, this is a good form here Apple has, but yeah.

**Wesley Hilliard** (1:31)
Yeah, but that was probably machine learning. Yeah, like actual generative AI, like actual. So when we say AI, it's a terrible name for a process. It's a large language model. It's basically the idea that was invented in the 80s, that we could teach computers to teach themselves, right?
More or less was the idea. That was the goal. And we, in order to make cars, level five autonomous vehicles, they need to think for themselves with the usual. I'm going to put the usual disclaimer here of computers can't think, AI doesn't think, it can't reason, it's not a person. I wish we would stop anthropomorphizing these things. But anyway, but to keep the language simple, because it's exhausting trying to go around it, we need better words. We wanted computers to think for themselves, so or at least be able to teach each other things, stuff like that. Self-driving car, obvious connection there.
Project Titan necessitated a lot of technology, which led us to the neural engine, the neural chip on Apple's silicon processors. So that led into more advanced machine learning, which was leaning into those decision trees that was being accelerated by a specific piece of hardware on the chip. And Apple was already getting into, what if we could have it go beyond a decision tree? It's just, ChatGPT came onto the scene in 2021, 2022 with its full LLM model, very early days. Think back to Will Smith eating spaghetti. It wasn't great.
But, okay, right.

**William Gallagher** (3:18)
I missed that episode, but all right, yeah, it's, don't look it up.

**Wesley Hilliard** (3:21)
It's horrifying. It's horrifying. AI generated video of Will Smith eating spaghetti.

**William Gallagher** (3:26)
So, oh, I see. Okay. Well, as long as it's an important use of generative AI. Yeah. Okay.

**Wesley Hilliard** (3:33)
So Apple's being told, well, you don't have the Will Smith spaghetti. What are you doing? So Apple in 2023 is, well, we have these projects, right? At that point, Apple had announced, I believe, correct me if I'm wrong, this is the year, WWDC 2023, was the year Apple came out on stage and said, autocorrect was shifting to a generative model basically. They use a very specific term and it's escaping me right now. But everyone's ears perked up and said, oh, that's AI, right? They're talking about it. So autocorrect is now AI, and it's the only thing on the iPhone that was technically generative. Then 2024 rolls around, and Apple tries to roll out its new thing with the hybrid system. We don't need to get into it. We all know it failed. Apple needed to backtrack and recreate it from the ground up for LLMs instead of trying to do a hybrid machine learning LLM.
In 2024, Apple started seeking out partnerships, and we know about this because of leaks and reports from places like Bloomberg.
Apple was courting everyone. They were talking to Anthropic, they were talking to Google, they were talking to OpenAI. At the time, the only one that said, yeah, we will let you use our LLM, no cost. All you have to do is kind of advertise it, you know, pay us with exposure, and you can use ChatGPT as an extension to Siri AI, so at the time, Apple Intelligence. And that's how they got into bed with each other. The problem is, it was already an unsteady relationship from the get go. Oh, Johnny Ive had already joined OpenAI to work on competitive hardware with Apple. The connection was tenuous at best. It was basically sharing data to a ChatGPT server with the promise that we won't collect anything, put into the contract, and it was only a two-year exclusivity contract, which means at the end of this year, we could see that expire and see more partnerships enter the extension. So in that time since, OpenAI has increasingly become adversarial with Apple, poaching over 400 members of its staff, high-profile hires, multiple engineers, high-level engineers and managers, on top of doing seedy things like working on killed drones for the US government. So a lot of weird stuff there.